The overarching goal of this project is to determine host epidemiologic and genetic factors that will be predictive of efficacy and toxicity of platinum-based chemotherapy or combined with thoracic radiotherapy in NSCLC patients. We will construct a well-characterized cohort of 1,200 NSCLC patients (Stages III and IV - receiving first-line platinum-based chemotherapy [unreadable] definitive thoracic radiotherapy). This cohort will then be studied for epidemiologic, clinical and a large number of rationally selected germline polymorphisms to correlate with the clinical outcome to allow us to construct predictive risk models for clinical efficacy and toxicity. We estimate there will be ~ 600 patients treated by platinum-based chemotherapy alone, and 600 Stage III patients receiving platinum-based chemotherapy plus definitive thoracic radiotherapy. There are three specific aims: 1) we will identify novel genetic loci that predict efficacy and toxicity to platinum-based chemotherapy and radiotherapy in all 1,200 patients. We will adopt a pathway-based genotyping and analyzing approach to evaluate frequencies .of about 8,000 SNPs in genes involved in pathways relevant to platinum and radiation response. We will examine individual SNP main effects, haplotypes, and the cumulative effect of SNPs in modulating efficacy and toxicity. Our hypothesis is that specific genotypes that alter the metabolism or action of platinum agents or relevant to the genotoxic effects of radiotherapy may impact the efficacy and toxicity of patients to these therapies. 2) we will apply machine-learning tools to identify gene-gene and gene-environment interactions influencing NSCLC outcome. We will develop algorithms to identify subgroups with differing platinum or radiotherapy treatment efficacy or toxicity. Our hypothesis is that therapeutic response is modulated by common, low penetrance polymorphisms, and that these polymorphisms interact with each other and/or host factors in determining response to therapy. 3) we will construct predictive risk models for survival and toxicity by integrating clinical and epidemiologic data with the genetic data from this project,and additional information from other R01 studies devoted to these cohorts such as a series of phenotypic assays. We hypothesize that the addition of genetic markers to the standard clinical and epidemiologic variables will improve the prediction of survival and toxicity of the final risk assessment models. We will compare the prediction accuracy among all patients, patients receiving chemotherapy alone, and patients treated by combined modality. The risk models resulting from this project may permit clinicians to identify patients before the start of therapy who are most and least likely to benefit or to develop toxicity and will have immense clinical benefit in terms of planning chemotherapy and radiotherapy for individual patients.